From: caleb
---
libavcodec/Makefile|2 +-
libavcodec/jpeg2000.h |3 +
libavcodec/jpeg2000dec.c | 69 +-
libavcodec/jpeg2000htdec.c | 1425
libavcodec/jpeg2000htdec.h | 28 +
5 files changed, 1511 insertions(+), 16 deletions(-)
crea
mån 2022-12-12 klockan 10:40 -0800 skrev Pierre-Anthony Lemieux:
>
> I suggest keeping optimizations as a second step, and instead
> focusing
> on accuracy of decoding so that we end-up with a good baseline.
I didn't check this yet since I'm busy with other projects, but if what
Caleb told me abo
On Fri, Dec 2, 2022 at 10:46 AM Tomas Härdin wrote:
>
> fre 2022-12-02 klockan 21:11 +0300 skrev etemesica...@gmail.com:
> >
> > +/**
> > + * Given a precomputed c, checks whether n % d == 0
> > + */
> > +static av_always_inline uint32_t is_divisible(uint32_t n, uint64_t
> > c)
> > +{
> > +re
Also thanks for reviewing
On Fri, Dec 2, 2022 at 10:34 PM Caleb Etemesi
wrote:
> Hi
>
> For bit-reading,
> 1. The spec has some chunks read from back to front,I didn't see that
> functionality present in get_bits.h(which I assumed contained the bit
> reading API).
> 2. It doesn't handle unstuffi
Hi
For bit-reading,
1. The spec has some chunks read from back to front,I didn't see that
functionality present in get_bits.h(which I assumed contained the bit
reading API).
2. It doesn't handle unstuffing correctly.
3. Doesn't handle EOB correctly, the spec has some arrays when there are no
more
fre 2022-12-02 klockan 21:11 +0300 skrev etemesica...@gmail.com:
>
> +/**
> + * Given a precomputed c, checks whether n % d == 0
> + */
> +static av_always_inline uint32_t is_divisible(uint32_t n, uint64_t
> c)
> +{
> + return n * c <= c - 1;
> +}
This looks like something that could go in la
From: caleb
This patch adds support for HTJ2K decoding in FFMPEG.
This is a revised patch with suggested changes from earlier and fixes some
bugs that were in the first version.
The paper describing HTJ2K can be found at
https://htj2k.com/wp-content/uploads/white-paper.pdf
---
libavcodec/Mak
